We have seen in the last year an incredible pace of progress in large AI models, with increasing abilities to generate high quality images, videos, text, sound and more. The best of these models display signs of creativity, reasoning, generalization and plasticity beyond what we could imagine just a few years ago. Yet many challenges and open questions remain, both on the technological aspects and the societal impact of these models. Further progress, especially on mitigating the social risks of these models, is hampered by a lack of transparency and reproducibility. In this talk, Joelle will describe ongoing efforts to increase best practices towards the responsible training and deployment of AI research systems, drawing on her experience with the ML reproducibility program, and the recent release of several state-of-the-art large models

Jennifer Doudna will discuss her professional and personal journey working on CRISPR technology, from its genesis to its applications today and focusing on ethical challenges that mirror challenges with AI/ML.

In this breakout session the leaders will discuss the following questions and challenges with attendees: ● What are the examples of systematic approaches for using generative and multi-modal models to evaluate and improve robustness of deep neural networks? ● What are the limitations of using generative models in terms of sample quality and diversity? ● What are the challenges with using generative modeling based tools in special domains (e.g. medical) where available data is more limited? ● How do we monitor the biases of generative and multi-modal models themselves?

Discussion of topics: a. Role of Mentorship: Empowering women, facilitating career development, building confidence, combating imposter syndrome and expanding professional networks. Give examples by personal stories. b. Seeking Guidance: Identifying potential mentors, contacting them. Guidance in the form of friendship - share personal stories. c. Maintaining Long-Term Professional Relationships: Regular communication, sharing updates on accomplishments, expressing gratitude. d. Networking: Through networking, women can gain visibility, create support systems (especially important as an immigrant), share experiences, exchange knowledge, and collaborate.

There are many robustness issues arising in a variety of forms when deploying ML systems in the real world. For example, neural networks suffer from sensitivity to distributional shift, when a model is tested on a data distribution different from what it was trained on. Such a shift is frequently encountered in practical deployments and can lead to a substantial degradation in performance. In addition, neural networks are vulnerable to adversarial examples - small perturbations to the input can successfully fool classifiers into making incorrect predictions. In this section, we will develop a deeper understanding of different robustness issues and discuss how to effectively enhance models' robustness.

We will have lunch and opportunities for networking in a round table format.

Round Table A: Apple -- Finding Mentors and Being a Mentor - Rishika Agarwal (Engineer) - Ivy Zhang (Engineer)

Round Table B: D. E. Shaw Research -- Machine Learning at D. E. Shaw Research - Jocelyn Sunseri (Machine Learning Research Engineer)

Round Table C: Google DeepMind -- Keeping Up With the Pace of Change in Industry - Kate Baumli (Research Engineer) - Kavya Kopparupu (Research Engineer)

Round Table D: Google Research -- Life and Work at Google - Alicia Parrish (Research Scientist, Responsible AI)

Round Table E: Microsoft -- Exploring Pathways: Career Opportunities, Growth, and Work-Life Balance at Microsoft Research - Lili Wu (Data and Applied Scientist, Microsoft Research) - Cyril Zhang (Senior Researcher, Microsoft Research)

Round Table F: Two Sigma -- Your Next Big ML Move: Innovation in Finance - Brittany Clarke (Diversity Recruiting Program Manager) - Alyssa Lees (Engineering Manager, News Engineering: a NLP Technology Team)

In the talk, Rihab will share her personal journey as a mid-career woman coming from Africa in the field of Artificial Intelligence (AI) and highlight the remarkable experiences she has gained working at an African AI startup. With a focus on both technical accomplishments and driving forces that have propelled her forward, she aims to inspire the audience while providing valuable insights into her professional growth - particularly to women who aspire to build their careers in AI.

Discussion Questions

1. How does data quality affect downstream trustworthy ML desiderata such as fairness, robustness, and explainability?
2. How do we measure data diversity and quantify the impacts of a lack of data diversity in downstream discriminative and generative tasks?
3. How do we weigh the importance of fairness and performance across training and deployment distributions in the presence of distribution shifts?
4. What common machine learning datasets are designed around features with may not predict the target label? How can we interrogate questions of predictability before developing algorithmic techniques for a particular task?

Discussion Questions

1. How does data quality affect downstream trustworthy ML desiderata such as fairness, robustness, and explain- ability?

2. How do we measure data diversity and quantify the impacts of a lack of data diversity in downstream discriminative and generative tasks?

3. How do we weigh the importance of fairness and performance across training and deployment distributions in the presence of distribution shifts?

4. What common machine learning datasets are designed around features with may not predict the target label? How can we interrogate questions of predictability before developing algorithmic techniques for a particular task?

There are many robustness issues arising in a variety of forms when deploying ML systems in the real world. For example, neural networks suffer from sensitivity to distributional shift, when a model is tested on a data distribution different from what it was trained on. Such a shift is frequently encountered in practical deployments and can lead to a substantial degradation in performance. In addition, neural networks are vulnerable to adversarial examples - small perturbations to the input can successfully fool classifiers into making incorrect predictions. In this section, we will develop a deeper understanding of different robustness issues and discuss how to effectively enhance models' robustness.